Introducing next-generation audio models in the API

For the first time, developers can also instruct the text-to-speech model to speak in a specific way - for example, “talk like a sympathetic customer service agent” - unlocking a new level of customization for voice agents.